The Net Stable Funding Ratio (NSFR) is one of two liquidity requirements included in the set of international bank regulation standards established by the Basel Committee on Banking Supervision (BCBS) following the 2007-09 financial crisis. The U.S. banking agencies put the NSFR out for public comment in April 2016 but have not adopted it. The Fed is under pressure from other countries on the Basel Committee to adopt the NSFR, which is an odd position for the Fed to be in, having implemented more-restrictive-than-required versions of most other Basel standards.
Adopting the NSFR at this time would be both ironic and reckless. As we discuss in this note, had the NSFR been in place in the United States, it would have made the September 2019 episode of repo market volatility and the March 2020 episode of Treasury market volatility — when the financial system walked up to the edge of catastrophe — worse. It would have done so because, despite being a liquidity requirement, the NSFR punishes banks for holding two of the most liquid investments available to banks: lending overnight with Treasury securities as collateral (aka “Treasury reverse repo”) and outright holdings of Treasury securities. Consequently, the NSFR would have added a further hindrance to banks providing the type of intermediation that was most needed during those events.
Moreover, after having so narrowly dodged the NSFR bullet over the past year, the Fed should not conclude that the regulation can just be tweaked and then adopted safely. It is not just a coincidence that BPI and many others (including the U.S. Treasury) have warned for years that the regulation will have costly unintended consequences. As BPI has documented extensively (see “Refurbishing the NSFR,” “Six Questions About The NSFR,” and “Send the NSFR Back to Basel” and the nine other notes cited therein), the regulation is fundamentally flawed. The NSFR was the last part of the original Basel III accord agreed upon by the BCBS, and its final version was a hastily thrown-together compromise designed to please too many masters. While the NSFR, in principle, serves an important purpose, it has significant structural problems and needs to be completely rethought. The banking agencies should not adopt it just to earn a higher mark in the annual Basel III monitoring report issued by the BCBS.
As a general matter, the NSFR is vaguely intended to ensure that a bank’s sources of funding are sufficiently stable to support its assets and activities over the course of one year. To that end, the NSFR is defined as the ratio of “available stable funding” (ASF) to “required stable funding” (RSF), and the largest banks would be required to maintain an NSFR ratio of at least 100 percent. ASF is determined by taking each bank liability and element of regulatory capital, multiplying it by an “ASF factor,” and then adding up all the resulting, weighted numbers. The ASF factors, which vary between 0 and 100 percent, are judgmental assessments of the reliability of each liability. RSF is similarly determined by summing bank assets weighted by “RSF factors.” The RSF factors are a judgmental assessment of the difficulty of liquidating an asset or the importance of maintaining a business line; like ASF factors, they also vary from 0 to 100 percent.
In mid-September 2019, a severe supply-demand imbalance in the Treasury repo market drove repo rates up from 2 percent to nearly 10 percent. Monday, September 16 was corporate tax day, and, as always, there were large outflows from money funds, which are important suppliers of repo financing. Simultaneously, $54 billion in Treasury securities settled, increasing the amount of securities in private hands that needed to be financed. The decline in repo funds available combined with the increase in demand for repo financing, occurring as it did against a backdrop of a declining but still a high level of reserve balances (banks’ deposits at the Federal Reserve), drove repo rates sharply higher on Monday and Tuesday, and markets remained fragile for a week.
The two big questions to come out of the episode was “Why didn’t banks respond to the higher rates by shifting funds out of their vast deposits at the Federal Reserve and into reverse repo?” and “Why didn’t banks borrow where it was cheaper and relend the funds into repo markets where rates had spiked, profiting from the spread?” Ultimately, the Fed decided that the answer to the former question is that banks want to hold a lot more reserves than had been thought, and the answer to the latter is that banks were reluctant to expand their balance sheet in part because of regulations. As discussed in Covas and Nelson (2019), any bank that borrowed and then relent into the repo market reduced its supplementary leverage ratio and its ability to pass the leverage ratio hurdles in the Fed’s annual stress tests and increased its GSIB surcharge.
If the NSFR had been adopted, the situation would have been worse. Both possible ways that a bank could have intermediated to equilibrate the supply-demand imbalance in repo markets would have caused the bank’s NSFR to fall. The NSFR does not require a bank to have stable funding for reserves, but it requires stable funding equal to 10 percent of Treasury reverse repos. As a result, every dollar a bank shifted from reserves to reverse repo would have raised its required stable funding by 10 cents, lowering its NSFR. Similarly, Treasury repo borrowing does not count at all as stable funding (it has an ASF of 0), so if the bank had lent funds into the repo market (with an RSF of 10 percent), received a Treasury security as collateral and used that Treasury security as collateral to borrow (a matched-repo transaction), the bank’s required stable funding would go up without any corresponding increase in its available stable funding, again driving down its NSFR.
It is notable that bank regulators and supervisors at the time pointed out that the Liquidity Coverage Ratio (LCR) and other bank liquidity requirements were designed to be unaffected when a bank shifted from reserves to reverse repo and so should not have inhibited banks and so should not have been contributors to the volatility in September. If the NSFR were adopted, as those same banking agencies have proposed, a bank’s liquidity requirements would be reduced when it made precisely that substitution.
In mid-March of this year, as the extent of the deterioration in the economic outlook became clear, there was a worldwide shift by investors into cash. To raise that cash, many investors sold Treasury securities. At the same time, many hedge funds dumped Treasuries as they unwound related futures positions that had moved against them. The resulting flood of Treasuries onto the market overwhelmed the capacity of broker-dealers, and market liquidity dried up. The two most commonly cited measures of market liquidity – market depth and bid-offer spread – both worsened by a factor of 10 (Duffie (2020)). The most liquid market in the world could have seized up entirely if the Federal Reserve had not purchased an extraordinary $1 trillion in securities over three weeks.
The breakdown in the Treasury market occurred because broker-dealers did not have the balance sheet capacity to absorb the massive sales. On March 16, when Vikram Rao, the head bond trader of Capital Group, asked executives that he knew at many of the big banks why they weren’t trading:
They had the same refrain: There was no room to buy bonds and other assets and still remain in compliance with tougher guidelines imposed by regulators after the previous financial crisis. In other words, capital rules intended to make the financial system safer were, at least in this instance, draining liquidity from the markets. One senior bank executive leveled with him: ‘We can’t bid on anything that adds to the balance sheet right now.’
Similarly, the Fed’s Monetary Policy Report to Congressin June 2020 stated: “However, in March, constraints on dealers’ intermediation capacity, including internal risk-management practices and regulatory constraints on the bank holding companies under which many dealers operate, were cited as possible reasons for deteriorating liquidity in even usually liquid markets.” Lorie Logan, executive vice president of the New York Fed stated “…facing balance sheet constraints and internal risk limits amid the elevated volatility, dealers had to cut back on intermediation.”
While the contribution of regulation to balance sheet constraints will no doubt continue to be debated, the proof is, to a great extent, in the pudding. To relieve strains on Treasury markets, the Federal Reserve soon thereafter excluded reserve balances and Treasury securities from bank holding companies’ supplementary leverage ratios (SLRs) – the regulation most frequently cited as the source of balance sheet constraints.
If the NSFR had been in place, the nearly catastrophic situation would have been even worse. While the LCR did not contribute to the March conflagration, the NSFR would have added fuel to the fire. If a broker-dealer had purchased a Treasury security and borrowed in the repo market to finance the purchase, its NSFR would have gone down. If it had lent to others to finance their purchases of Treasury securities and borrowed in the repo market to finance that lending, its NSFR would have gone down.
In addition, some of the elements included in the U.S. NSFR proposal but not in the Basel standard would also have worsened the situation in March. In particular, the heightened market volatility experienced in March and April increased significantly the market value of derivative liabilities. This volatility would have increased the required stable funding requirements for gross derivative liabilities, driving down the NSFR.
Both September and March
In neither episode of volatility was any large financial institution faced with funding problems. That is, while the events were replete with problems that would have been made worse by the NSFR, there were no problems that would have been made better, and thus no evidence the NSFR was needed. Speaking in terms of a cost-benefit analysis, the costs would have been extremely high and the benefits zero.
The NSFR needs to be rethought
Even though the stable funding requirements in the NSFR on repos and Treasuries are the flaws that would have made financial markets less liquid and the volatility in September 2019 and March 2020, the banking agencies cannot just tweak those aspects of the regulation and safely adopt the NSFR.
The NSFR is chock full of flaws, mostly ones introduced when the regulation rushed to completion. Four years ago, in “The Net Stable Funding Ratio, Neither Necessary nor Harmless,” we presented reasons why the NSFR may reduce lending to nonfinancial businesses and households, reduce financial resilience, degrade financial market functioning, judge a financial institution illiquid that is in fact fine, and judge and institution fine that is, in fact, illiquid. The NSFR serves an important purpose – the 30-day horizon of the LCR is too short for the regulation to stand alone. But there are better ways to build a liquidity regulation that avoids cliff effects (see here), and there are ways to calibrate the NSFR to actual data (see here). It would be reckless to simply make small changes to the NSFR and adopt it. As proposed in the United States, it is capricious and needs to be rethought.
As a final note, the Fed has often said that it does not adopt Basel standards simply because the BCBS has agreed to them. Rather, it only implements those whose benefits outweigh their costs in the United States. For example, in Congressional testimony in November 2015, then Chair Janet Yellen said:
[The BCBS] is a very useful committee. We participate actively. We want to make sure that other countries put in place tough safety and soundness regulations that will be good for our firms and for financial stability. But nothing is law in the United States or is adopted as a regulation unless we deem it to be appropriate for our firms, and I believe all countries behave in the same manner. These international bodies are coordinating bodies where consultation takes place, but that doesn’t substitute for domestic rule-writing efforts here in the United States.
But as we describe in “Foundational Basel Committee Study Estimates the Costs of NSFR Exceed its Benefits by Nearly $1 Trillion,” when the Fed proposed the NSFR, it cited a 2010 Basel study (available here) as demonstrating that the benefits of the NSFR exceeded its costs. However, the Fed misread the results of the study. At current capital levels, the Basel study found that the costs of the NSFR exceed its benefits. By its own criteria, the Fed should withdraw its proposal and undertake to redesign the NSFR in Basel. Moreover, the failed cost-benefit test doesn’t account for the harm that the NSFR would do to market-based finance, for which well-functioning repo markets are critical. Consequently, even if the NSFR were appropriate in an economy with a bank-centric financial system, it is wholly inappropriate in the United States, where market-based finance is so important.
It seems certain that the effect of NSFR adoption over the long term will be to cause banks to increase further the shares of their assets that are unproductive government securities and deposits at the Fed and reduce shares that are loans to households and non-financial businesses or reverse repos that support market-making activity. That seems an odd choice in the middle of an episode of market stress that has demonstrated that banks have ample liquid assets on their books, that lending to households and businesses is critical for bridging the pandemic disruption, and that market-making activity has been so disadvantaged and therefore deemphasized that the Federal Reserve has had to intervene as buyer-of-last-resort across all fixed income markets from repo to municipals to corporate debt. Sometimes, what happens in Basel should stay in Basel.
Those of us working on the Basel liquidity requirements were focused on finishing the Liquidity Coverage Ratio, and we often joked that “one day we really have to return to fixing the problems in the NSFR”.
The two forms of funding that count the most toward stable funding are equity and long-term debt. Consequently, rough intuition about the impact of the NSFR on repo and Treasury market liquidity can be obtained by thinking of the proposed 10 and 5 percent stable funding requirements for reverse Treasury repos and Treasury securities, respectively, as 10 and 5 percent capital requirements on those two assets.
See “The Economic Outlook, Monetary Policy, and the Demand for Reserves,” Speech by Vice Chair for Supervision Randal K. Quarles, February 6, 2020. https://www.federalreserve.gov/newsevents/speech/quarles20200206a.htm.
In fact, by design, the LCR is not changed by any short-term lending or borrowing between financial institutions, regardless of whether or not the transaction is collateralized and if so, with what type of collateral. As a result, the LCR does not build in an incentive for banks to pull away from each other when markets are stressed. The NSFR, by contrast, does.
Justin Baer, “The Day Coronavirus Nearly Broke the Financial Markets,” Wall Street Journal, May 20, 2020.
Although the Fed did not exclude reverse Treasury repos from the SLR, the exclusion of Treasury securities and reserve balances reduced the extent to which the SLR was binding, easing the capital costs of those reverse repos.
In the Basel Committee’s October 2011 report to the G20, the Committee indicated “…members agreed to implement Basel III from 1 January 2013” (see pp. 1-2 of the progress report here), but it failed to meet that deadline in the case of the NSFR. The BCBS announced on January 6, 2013 that final changes to the LCR had been agreed and that it will “…press ahead with the NSFR…” (see press release here). In their progress report to the G20 in April 2013, the BCBS indicated that the NSFR should be completed in 2014. The BCBS issued the final standard for the NSFR on 31 October 2014 (see press release here), just at the last minute to make the BCBS’s progress report to the G20 on 12 November 2014 (see here) in which it stated that it had “…largely completed its post crisis reform agenda….,” avoiding having missed its own deadline yet again.
That said, in the United States large banks already comply with enhanced prudential standards requiring monthly liquidity stress tests looking at a horizon of overnight, 30 days, 90 days and one year, with certain smaller institutions (referred to as Category IV banks) performing such stress tests on a quarterly basis.